GONRG-2147: Fix double colon in refs

src/dags/libs/manifest_analyzer.py

@@ -97,8 +97,11 @@ class EntityNode(object):
         content = json.dumps(self.content)
         for parent in self.parents:
             if parent.system_srn:
-                # ':' at the end is for showing that it is reference.
-                content = content.replace(parent.srn, f"{parent.system_srn}:")
+                # ':' at the end is for showing that it is reference if parent srn was surrogate-key.
+                if "surrogate-key" in parent.system_srn:
+                    content = content.replace(parent.srn, f"{parent.system_srn}:")
+                else:
+                    content = content.replace(parent.srn, f"{parent.system_srn}")
         self.content = json.loads(content)
 
 

tests/plugin-unit-tests/test_manifest_analyzer.py

@@ -86,7 +86,10 @@ class TestManifestAnalyzer(object):
         return manifest_analyzer
 
     def process_entity(self, entity: EntityNode) -> str:
-        return f"system_srn: {entity.srn}"
+        if "surrogate-key" in entity.srn:
+            return f"system_srn: {entity.srn}"
+        else:
+            return entity.srn
 
     def index_in_queue_by_srn(
         self,
@@ -200,3 +203,50 @@ class TestManifestAnalyzer(object):
         logger.info(f"Processed entities: {manifest_analyzer.processed_entities}")
         logger.info(f"Unprocessed entities {manifest_analyzer.unprocessed_entities}")
         assert "surrogate-key:wpc-1" in (e.srn for e in manifest_analyzer.unprocessed_entities)
+
+    @pytest.mark.parametrize(
+        "manifest,expected_replaced_srns",
+        [
+            pytest.param(
+                [
+                    {
+                        "id": "surrogate-key:wpc",
+                        "ref": "surrogate-key:wpc2"
+                    },
+                    {
+                        "id": "surrogate-key:wpc2"
+                    }
+
+                ],
+                "system_srn: surrogate-key:wpc2:",
+                id="Surrogate key"
+            ),
+            pytest.param(
+                [
+                    {
+                        "id": "surrogate-key:wpc",
+                        "ref": "osdu:reference-data--ResourceSecurityClassification:RESTRICTED:test:"
+                    },
+                    {
+                        "id": "osdu:reference-data--ResourceSecurityClassification:RESTRICTED:test"
+                    }
+
+                ],
+                "osdu:reference-data--ResourceSecurityClassification:RESTRICTED:test:",
+                id="Real Ids"
+            )
+        ]
+    )
+    def test_update_parent_srn_with_system_srn(self, manifest, expected_replaced_srns):
+        data = [ManifestEntity(entity=e, schema="") for e in manifest]
+        token_refresher = AirflowTokenRefresher()
+        manifest_analyzer = ManifestAnalyzer(
+            data,
+            token_refresher
+        )
+
+        for entity in manifest_analyzer.entity_queue():
+            entity.replace_parents_surrogate_srns()
+            entity.system_srn = self.process_entity(entity)
+            if entity.content.get("ref"):
+                assert entity.content["ref"] == expected_replaced_srns